A GENERALIZATION OF STIRLING NUMBERS

Similar documents
SOME IDENTITIES INVOLVING THE FIBONACCI POLYNOMIALS*

Some Congruences for the Partial Bell Polynomials

Wilson s Theorem and Fermat s Little Theorem

ON A KIND OF GENERALIZED ARITHMETIC-GEOMETRIC PROGRESSION

A Symbolic Operator Approach to Several Summation Formulas for Power Series

CONGRUENCES FOR BERNOULLI - LUCAS SUMS

On the Convergence of the Summation Formulas Constructed by Using a Symbolic Operator Approach

On the indecomposability of polynomials

w = X ^ = ^ ^, (i*i < ix

±i-iy (j)^> = P"±(-iy (j)#> (2)

SOME IDENTITIES INVOLVING GENERALIZED GENOCCHI POLYNOMIALS AND GENERALIZED FIBONACCI-LUCAS SEQUENCES

Here, logx denotes the natural logarithm of x. Mrsky [7], and later Cheo and Yien [2], proved that iz;.(»)=^io 8 "0(i).

Review Problems for Midterm Exam II MTH 299 Spring n(n + 1) 2. = 1. So assume there is some k 1 for which

*!5(/i,*)=t(-i)*-, fty.

Arithmetic properties of lacunary sums of binomial coefficients

Takao Komatsu School of Mathematics and Statistics, Wuhan University

F. T. HOWARD AND CURTIS COOPER

ON DIVISIBILITY OF SOME POWER SUMS. Tamás Lengyel Department of Mathematics, Occidental College, 1600 Campus Road, Los Angeles, USA.

Another Proof of Nathanson s Theorems

1. Introduction Definition 1.1. Let r 1 be an integer. The r-generalized Fibonacci sequence {G n } is defined as

(n = 0, 1, 2,... ). (2)

Congruences for central factorial numbers modulo powers of prime

p-regular functions and congruences for Bernoulli and Euler numbers

Math 0320 Final Exam Review

Explicit formulas for computing Bernoulli numbers of the second kind and Stirling numbers of the first kind

= (w 0 + w x t Up^t 9 ' 1 )-. A. "0

SOME AMAZING PROPERTIES OF THE FUNCTION f(x) = x 2 * David M. Goldschmidt University of California, Berkeley U.S.A.

Mathematical Olympiad Training Polynomials

On the power-free parts of consecutive integers

Mathematica Slovaca. Jaroslav Hančl; Péter Kiss On reciprocal sums of terms of linear recurrences. Terms of use:

A RESULT ON RAMANUJAN-LIKE CONGRUENCE PROPERTIES OF THE RESTRICTED PARTITION FUNCTION p(n, m) ACROSS BOTH VARIABLES

arxiv: v1 [math.co] 21 Sep 2015

THE p-adic VALUATION OF EULERIAN NUMBERS: TREES AND BERNOULLI NUMBERS

On Arithmetic Properties of Bell Numbers, Delannoy Numbers and Schröder Numbers

Part II. Number Theory. Year

Y. H. Harris Kwong SUNY College at Fredonia, Fredonia, NY (Submitted May 1987)

F O R M U L A S F O R CONVOLUTION F I B O N A C C I NUMBERS AND P O L Y N O M I A L S

Number Theory Homework.

fn" <si> f2(i) i + [/i + [/f] (1.1) ELEMENTARY SEQUENCES (1.2)

MULTISECTION METHOD AND FURTHER FORMULAE FOR π. De-Yin Zheng

arxiv: v1 [math.nt] 8 Jan 2014

POLYNOMIAL SOLUTIONS OF PELL S EQUATION AND FUNDAMENTAL UNITS IN REAL QUADRATIC FIELDS

ON THE EQUATION Xa=7X+/3 OVER AN ALGEBRAIC DIVISION RING

arxiv: v2 [math.nt] 4 Jun 2016

ASYMPTOTIC DISTRIBUTION OF THE MAXIMUM CUMULATIVE SUM OF INDEPENDENT RANDOM VARIABLES

FIBONACCI NUMBERS AND SEMISIMPLE CONTINUED FRACTION. and Ln

PROBLEMS ON CONGRUENCES AND DIVISIBILITY

Review problems solutions

ON VALUES OF CYCLOTOMIC POLYNOMIALS. V

Definitions & Theorems

Introduction to Lucas Sequences

ADVANCED PROBLEMS AND SOLUTIONS

FUNCTIONS OVER THE RESIDUE FIELD MODULO A PRIME. Introduction

MULTI-RESTRAINED STIRLING NUMBERS

Today. Polynomials. Secret Sharing.

Applications. More Counting Problems. Complexity of Algorithms

Acta Mathematica Universitatis Ostraviensis

Some congruences for Andrews Paule s broken 2-diamond partitions

1. INTRODUCTION. n=. A few applications are also presented at the end of the Note. The following Theorem is established in the present Note:

Degenerate Bernoulli polynomials, generalized factorial sums, and their applications

A q-analogue OF THE GENERALIZED FACTORIAL NUMBERS

Proof 1: Using only ch. 6 results. Since gcd(a, b) = 1, we have

PUTNAM TRAINING NUMBER THEORY. Exercises 1. Show that the sum of two consecutive primes is never twice a prime.

A LINEAR BINOMIAL RECURRENCE AND THE BELL NUMBERS AND POLYNOMIALS

A CRITERION FOR POLYNOMIALS TO BE CONGRUENT TO THE PRODUCT OF LINEAR POLYNOMIALS (mod p) ZHI-HONG SUN

Lagrange s polynomial

Generating Functions (Revised Edition)

On Certain Sums of Stirling Numbers with Binomial Coefficients

Department of Mathematics, Nanjing University Nanjing , People s Republic of China

E.J. Barbeau. Polynomials. With 36 Illustrations. Springer

ON EXPLICIT FORMULAE AND LINEAR RECURRENT SEQUENCES. 1. Introduction

Lagrange s polynomial

Math 314 Course Notes: Brief description

Congruence of Integers

THE LOG-BEHAVIOR OF THE SEQUENCE FOR THE PARTIAL SUM OF A LOG-CONVEX SEQUENCE. 1. Introduction

Lawrence Somer Department of Mathematics, Catholic University of America, Washington, D.C (Submitted February 2001) 1. I N T R O D U C T I O N

THE DENOMINATORS OF POWER SUMS OF ARITHMETIC PROGRESSIONS. Bernd C. Kellner Göppert Weg 5, Göttingen, Germany

Successive Derivatives and Integer Sequences

be a formal power series with h 0 = 1, v(h^) = \ and v(h n ) > 1 for n > 1. Define a n by

On arithmetic functions of balancing and Lucas-balancing numbers

A Few Elementary Properties of Polynomials. Adeel Khan June 21, 2006

VAROL'S PERMUTATION AND ITS GENERALIZATION. Bolian Liu South China Normal University, Guangzhou, P. R. of China {Submitted June 1994)

(308 ) EXAMPLES. 1. FIND the quotient and remainder when. II. 1. Find a root of the equation x* = +J Find a root of the equation x 6 = ^ - 1.

Transformation formulas for the generalized hypergeometric function with integral parameter differences

Two Identities Involving Generalized Fibonacci Numbers

MATH39001 Generating functions. 1 Ordinary power series generating functions

An integer p is prime if p > 1 and p has exactly two positive divisors, 1 and p.

NEW GENERATING FUNCTIONS FOR CLASSICAL POLYNOMIALS1. J. w. brown

64 P. ERDOS AND E. G. STRAUS It is possible that (1.1) has only a finite number of solutions for fixed t and variable k > 2, but we cannot prove this

Part IA Numbers and Sets

THE MODULAR CURVE X O (169) AND RATIONAL ISOGENY

SECOND-ORDER RECURRENCES. Lawrence Somer Department of Mathematics, Catholic University of America, Washington, D.C

SYMMETRY AND SPECIALIZABILITY IN THE CONTINUED FRACTION EXPANSIONS OF SOME INFINITE PRODUCTS

Algebra Review 2. 1 Fields. A field is an extension of the concept of a group.

ADVANCED PROBLEMS AND SOLUTIONS

Course MA2C02, Hilary Term 2013 Section 9: Introduction to Number Theory and Cryptography

Know the Well-ordering principle: Any set of positive integers which has at least one element contains a smallest element.

Colloq. Math. 145(2016), no. 1, ON SOME UNIVERSAL SUMS OF GENERALIZED POLYGONAL NUMBERS. 1. Introduction. x(x 1) (1.1) p m (x) = (m 2) + x.

Representing numbers as the sum of squares and powers in the ring Z n

A Family of Sequences Generating Smith Numbers

Transcription:

Hongquan Yu Institute of Mathematical Sciences, Dalian University of Technology, Dalian 6024, China (Submitted September 996-Final Revision December 996). INTRODUCTION Let W(x), fix), g(x) be formal power series with complex coefficients, W(x) * 0, W{0) =, /(0) = g(0) = 0, Then the coefficients {B^n, k), B 2 (n, k)} in the following expansions, W(x)<J(x)f lk\ = B f a k ) x " l n \, (g{x)f /[W(g(x))k\] = ^B 2 (n,k)x n ln\, () ri>k are called a weighted Stirling pair if f(g(x)) = g(f(x)) = x, i.e.,/ g are reciprocal. When W(x) = l, B^in.k) B 2 (n,k) reduce to a Stirling type pair whose properties are exhibited in [7]. In this paper, we shall present a weighted Stirling pair that includes some previous generalizations of Stirling numbers as particular cases. Some related combinatorial arithmetic properties are also discussed. 2. A WEIGHTED STIRLING PAIR Let t, a, P be given complex numbers with a-(*0. Let f(x) = [( + ax)p' a -]//?, g(x) - [(\+px) alp -l]la, W(x) = (l + ax) t/a. Then, in accordance with (), by noting that f(x) g(x) are reciprocal, we have a weighted Stirling pair, denoted by {S(n, k, a, /?; t\ S(n, k, /?, a, -)} = { >, k), B 2 (n, k)}. We call it an (a, /?; /) [resp. a (ft, a; -t)] pair for short. Moreover, one of the parameters a or p may be zero by considering the limit process. For instance, a (,0; 0) [resp. a (0,; 0)] pair is just Stirling numbers of the first second kinds. Note that from the definition of an (a, /?; t) pair the first equation in (), we may obtain the double generating function of S(n, k, a, /?; t) as (l + ax)' /a cxpl (l + ax^a~ l \ = ^S(n,k,a,B-, t)^~u k. (2) If we differentiate both sides of (2) on x, then multiply by (l + ax) compare the coefficients of x n u k, we have S(n, k-l,a,ftt+p) = S(n +, k, a, fi; t) + (na - t)s(n, k, a, /?; t\ (3) if we differentiate both sides of (2) on u then compare the coefficients of x n u k:, we have S(n y k,a,p;t+p) = P(k + l)s(n,k + \a,p;t) + S(n,k,a,P;t). (4) Thus, the recurrence relation satisfied by S(n, k, a, /?; t) may be obtained by combining (3) (4): S(n + \ k, a,p; t) = (t + fik-ari)s(n, k, a,/3; t) + S(n,k-l, a,p, t). (5) 252 [JUNE-JULY

The initial values of S(n, k, a, /?; t) may be verified easily from () because S(n, 0, a, J3; t) = t(t-a)(t-2a) - (t-(n-l)a) for/i>l, S(n,n,a,fi;t) = l for n>0, S(n, k, a, fi;t) = 0 for k>n. Thus, a table of values of S(n, k, a, J3; t) can be given by concrete computations. «" 0 2 3 TABLE. S(n 2t + /}-a (t+fi-2a) +, K, t(t--aa,fi;t) 0 t t(t- -a) t(t- -a) 2 3 3t + 30-3a From (2), we may get the explicit expression for S(n, k, a, J3; t) via the generalized binomial theorem along the lines of (4.) in [6]. For a complex number a, define the generalized factorial of x with increment a by (x\a) n - x(x - a)(x -2a) - - - (x -na + a) for w =,2,..., (x\a) 0 =. Theorem : The (a, /?; t) pair defined by () may also be defined by the following symmetric relations: ((x + /) a) = S(n, K a, ft; t)(x\fi) k ; fc=0 (*!/?) = s(«,*, A «;-0((*+0l«)*- k=0 Proof: The proof of the theorem may be carried out by the same argument used by Howard [6], by showing that the sequences defined by (6) (7) satisfy the same recurrence relations have the same initial values as that of an (a, /?; t) pair. D Examples: Let A, 0 * 0 be two complex parameters. The so-called weighted degenerate Stirling numbers (S x (n, k,a\0),s(n, k, X\0)) were first introduced discussed by Howard [6] with definitions ( l - x ) - ^ " 0 - *>*! =k\y d S l {n,k,x\0)^ V * ) " {\+ex)^n\+exy -\f ^k^si^kx^^-, riik n\ where 9p = \. Now it is clear that (-l)"-^,(»,k, \,X\0) = S{n,k,\,9;9-X) S{n,k,X\6) = S{n,k,6,\;X). The limiting case 0 = 0, X * 0, gives the weighted Stirling numbers (R^n, k, X), R 2 (n, k, X)) discussed by Carlitz ([2], [3]) with definitions (6) (7) (l-xy\-log(\-x)) k = *! *,(», k, X)?- n ' 998] 253

where the weight function et* comes from the limit of ( + 0t) xie as 0 -> 0. It is apparent that ((-l)^%(n, Jfc, X\ R 2 (n, k, X)) forms a (,0; - X) pair. Further examples are the degenerate Stirling numbers [] defined by H - ((i+ftr-i)* = *!2>i,* 0), m where 0// =. It is clear that {(-\) n - k S x {n, k\0\ S(n, k\0)) is a (,0;0) pair. The noncentral Stirling numbers were first introduced by Koutras in [8] with the definitions: (0 = 5> a (»,*X'-«) k ; (t-a)" = ^S a (n,k)(t) k. It is now clear by Theorem that (s a (n, k), S a (n, k)) is a (,0; a) pair. Jc=0 3. REPRESENTATIONS OF WEIGHTED STIRLING PAIRS For r > 0, f r * 0, let F(x) = Z* =r f k x k I k! JF(x) = T% WjX j I j! be two formal power series. Following Howard [6], for complex z, we define the weighted potential polynomial F k (z) by ^l^f" fc=0 '*-*'-* =< F k (zyx*/k\. (8) Moreover, if r >, define the weighted exponential Bell polynomial B n k (0,..., 0, / r, / r+,...) by *P(x)[F(x)]* =.*!fx,(0,..., 0/ r,/ r +,...)*" ln\. (9) The following lemma is due to Howard ([6], Th. 3.). Lemma 2: With F k (z) B nk defined above, we have Now, from (9) with W(x) = ( + ox:)' 7 " F(x) = [( + ax) /3,a -] / /?, we have S{n 9 k,a 9 frt) = B^k(\p-a 9 tf-a)<fi-2a\w^ (0) Define the weighted potential polynomials A k (z) by 254 [JUNE-JULY

If we differentiate both sides of () with respect to x, then multiply by l + ax compare the coefficients of x k, we obtain It follows that with initial conditions za k {z +) = (z - k)a k {z) + k(t + (a - fi)z -(k- \)d)a k _ x {z). (-l) k [ k ~r i y k (rr + l) = (-d k { k ~ k "y k (n)ht + (a-fi) n -(k-vax-lf-^-^yu"), ^~" 0 " J4 ) («+ l) = l,for»>0, (3) (-!)"(~^\A n (n + \) = (t + a-p){t + a-2p)---{t + a-np), for n>\. (4) Therefore, by equations (2)-(4), the recurrence relations satisfied by S(n,n-k,fi, a; t + a- fl) [may be deduced from (5)] its initial values, we have that {_ l f(k-n-^ak{n + \) = S{n,n-k,p,a-t + a-p). It then follows from Lemma 2, by taking r = (0) that (2) By symmetry, we have the following representation formulas for weighted Stirling pairs. Theorem 3: For S(n 9 k 9 a, /?; t) defined by () S(n, k,f3,a;t + a-{t) defined in a like way, we have (»,*, a, 0 = l V l ) { ^ J ^ ( 5 > S(n, k,fi,cr,t + a-fl = %(-&$-k*}){ ^ ( 6 > Remark: It should be pointed out that similar representation results for the particular case when a = 0, 0=, t = l-x has been proved by Howard [6]. Here we borrow his proof techniques. 4. CONGRUENCE PROPERTIES OF WEIGHTED STIRLING PAIRS A formal power series $(x) = E w > 0 ^ " /«! is called a Hurwitz series if all of its coefficients are integers. It is well known that, for the Hurwitz series flx) with a Q = 0, the series {(f>{x)) k lk\ is again a Hurwitz series for any positive integer k. 998] 255

In this section we always assume a,/3,t e Z. Then it is clear that both (f(x)) k /k\ (g(x)) k /k\ in () are Hurwitz series, so that S(n, k, a, J3; t) S(n, k, /?, a; -i) are two integer sequences. First, let t = 0. Then we have Theorem 4: Let p be a prime number let k y be integers such that j + < k < p. Then the following congruence relation holds: S(p + j,k,p,a;0)^0(modp). (7) Proof: Assume first that a 4 0 (mod /?). For a polynomial ^(x) of degree n in x, we may express it, using Newton's interpolation formula, in the form <f>(x) = <f>(a 0 ) + k=\ y [a 0 a l,..a k ]{x\a} k, where [a 0 a l...a k ] denotes the divided difference at the distinct points x = a Q, a l?..., a k9... {x a) k =(x- a 0 )(x - a x ) (x - a k _ x ). Moreover, we have [a 0 a l...a k ] = a 0 a x CLu a\ l <f>(a Q ) a k ~ l Mad <4 <l>(a k ) a 0 a x CCu k-i o k- «0 of a * a> Now take (/> p {x) = (x\fi) p, then ^,(0) = 0. We have, by (7) (8), that (8) (9) 5(p,*,/?,a;0) = [a 0 a...aj, (20) which may be expressed as a quotient of two determinants as in (9), where a^ - ja (j = 0,,2,...). Notice that the classical argument of Lagrange that applied to the proof of may also be applied to prove the relation (x-l)->(x-p + l) = x p - l -l (mod/?) <f> p (x) = (x\p) p = x(x-fl)...( x -(p-i)0) S x'-fir-ix (mod/7), (2) where the congruence relation between polynomials are defined as usual (cf. [4], pp. 86-87, Th. 2). Also, using Fermat's Little Theorem, we find WaXJaY-fi^Ua)* \ja (mod/?), if/? /?, 0 (mod/?), ifp\p, where j = 0,,2,... Consequently, we obtain, witha^ = ja for k >, Moreover, the denominator is given by a 0 a k ~ l j p (i a x a\~ l <j> p {a x )\ = 0 (mod/?). a k - <A~ l t P (<* k ) 256 [JUNE-JULY

a la ka a k {kaf~ l a k {kaf = a' Y\(j-/) # 0 for k < p (mod/?). 0<i<j<k Thus, we have that S(p, k, fi, oc; 0) = 0 (mod /?) for < k < p. Furthermore, let F(x) = (x \fi) p+j. We then have F(x) =! / S(p+j, k, fi, a; 0)(x a) k F(x) = 0 p (x)(x-pfi)..-(x-(p + j)fi+fi) ^ (xp - pp- l x)x(x - fi)... (x - (J - \)fi) (mod /?) = (x p - fi p ~ l x){x j + ap s - x + + a^x) (mod /?), (22) where a l9..., a j-x e Z. Consequently, we have, for < i < p + y, F(ia) = (mod/>), if Pi A (ia) y+ + ^(/a) 7 + + aj^qaf (mod/?), if/?\fi. Since j < k -, we have a 0 aj,fc-l 0 F(a 0 ) F(a,) = 0 (mod/?), a t a,k-i where the last column is a linear combination of the first k columns modulo/?. Again, the same denominator determinant is not congruent to zero modulo /? for k <p. Thus, we have that S(p + j, k, fi, a; 0) = 0 (mod/?) for j + < k < p. The case for a = 0 (mod/?) may be proved directly using (7), (2), (22) by comparing the corresponding coefficients of powers of x in both sides of (2) (22). Hence, the theorem is proved. Note that in the particular case in which a =, fi = 0 or fi=, a = 0, Theorem 4 reduces to congruences for Stirling numbers of the first second kinds; see [5] for other congruences for Stirling numbers. Corollary 5: Let a, fi, t be integers. Then the (a, fi; t) pair satisfies the basic congruence S(p, k, a, fi;t) = 0 (mod p), (23) where/? is a prime \<k<p. Proof: Let W(x) = (l + ax) t/a = I dn > 0 a rj x n /n\ with a n ez,a 0 = l. Then it is clear from () that J]S(n,k,a,fi;t)x»M = \^a n xvn\ \Ws(n,k,a,fi;0)x«ln\ «>o so that we have S(p, k, a, fi; t) = f.a^sq, k, a, fi; 0){ p \ i=k V / 998] 257

From Theorem 4 (taking j = 0) the fact that (?) = 0 (mod p) for 0 < i < p, it follows that S(p, k, a, /?; r) = 0 (mod/?), the corollary is proved. ACKNOWLEDGMENT The author is very grateful to the referee for his valuable suggestions which considerably improved the presentation of this article. REFERENCES. L. Carlitz. "Degenerate Stirling, Bernoulli Eulerian Numbers." UtilitasMath. 5 (979): 5-88. 2. L. Carlitz. "Weighted Stirling Numbers of the First Second Kind-L" The Fibonacci Quarterly 8.2 (980): 47-62. 3. L. Carlitz. "Weighted Stirling Numbers of the First Second Kind-II." The Fibonacci Quarterly 8.3 (980):242-57. 4. G. H. Hardy & E. M. Wright. An Introduction to the Theory of Numbers. Oxford: Oxford University Press, 98. 5. F. T. Howard. "Congruences for the Stirling Numbers Associated Stirling Numbers." Acta Arithmetica 55 (990):29-4. 6. F. T. Howard. "Degenerate Weighted Stirling Numbers." Discrete Math. 57 (985):45-58. 7. L. C. Hsu. "Some Theorems on Stirling-Type Pairs." Proceedings of Edinburgh Math. Soc. 36 (993):525-35. 8. M. Koutras. "Non-Central Stirling Numbers Some Applications." Discrete Math. 42 (982):73-89. AMS Classification Numbers: 05A5, B73, A07 NEW EDITOR AND SUBMISSION OF ARTICLES Starting March,998 9 all new articles must be submitted to: PROFESSOR CURTIS COOPER Department of Mathematics Computer Science Central Missouri State University Warrensburg, MO 64093-5045 e-mail:' cnc885@cmsu2xmsu.edu Any article that does not satisfy all of the criteria as listed on the inside front cover of the journal will be returned immediately. [JUNE-JULY